No — not on this device
Nemotron 3 Super 120B-A12B at Q4_K_M needs 70.4 GB against 10.6 GB usable, and the shortfall of 59.8 GB is more than 32 GB of system RAM can cover at a tolerable speed. A smaller sibling or a lower quantisation is the honest answer here.
Does not fit
8K context
Q4_K_M · 69.7 GB
NVIDIA Open Model
Released Mar 2026
Not in the Ollama library
The open-training-data 120B. Same hybrid layout as Lightning, so 128K context costs under a gigabyte.
The VRAM budget
weights 69.7 GB
Weights 69.7 GB
KV cache @ 8K 0.06 GB
Runtime overhead 0.6 GB
Over budget 59.8 GB past 10.6 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 122.7 GB | 123.4 GB | — | ~1.3 | −0.1% ppl | 112.8 GB over |
| Q6_K | 94.7 GB | 95.4 GB | — | ~1.7 | −0.4% ppl | 84.8 GB over |
| Q5_K_M | 81.8 GB | 82.5 GB | — | ~2.0 | −0.8% ppl | 71.9 GB over |
| Q4_K_M | 69.7 GB | 70.4 GB | — | ~2.4 | −1.9% ppl | 59.8 GB over |
| Q3_K_M | 56.4 GB | 57.1 GB | — | ~3.1 | −5.4% ppl | 46.5 GB over |
| Q2_K | 48.4 GB | 49.0 GB | — | ~3.7 | −15% ppl | 38.4 GB over |
Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. Only 8 of its 88 blocks keep a per-token KV cache; the rest are linear-attention, Mamba or convolution blocks with a fixed-size state.
How to run it
$ llama-server \
-hf nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16:Q4_K_M \
-c 8192 -ngl 12
The engine underneath most of the others. Every knob is exposed. More on llama.cpp.
01Download is 69.7 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02Close anything else holding VRAM. A browser with hardware acceleration can sit on 1–2 GB.
03If it falls back to CPU silently, drop the context first, then step down a quantisation.